This Article was published in Harvard Business Review.
Why Ontology Is the Missing Foundation for AI Success: Insights From Seth Earley in Harvard Business Review
Harvard Business Review featured an article by Seth Earley on why so many enterprise AI initiatives stall despite large investments, and why the solution begins with a disciplined information foundation. Seth explains that companies continue to struggle with AI because the underlying data that powers these systems is fragmented, inconsistent, and poorly structured.
“Promises of AI vendors do not pay off unless a company’s data systems are properly prepared for AI,” Seth notes in the interview. He emphasizes that without clear structure across systems, AI cannot access the information it needs to generate reliable results.
Why AI Initiatives Break Down
Seth highlights the core reason behind failed AI programs: most organizations do not treat knowledge as an enterprise asset. Data remains locked in silos, disconnected across applications, and out of sync with business processes. As a result, even well funded AI projects deliver inconsistent outputs and fail to scale beyond pilots.
According to Seth, the real barrier is not the AI itself, but the lack of a unified information model that ties the organization’s data together in a consistent, accessible way.
Ontology as the Foundation for AI
Seth describes ontology as the master representation of all data, relationships, and concepts across the business. It is the organizing framework that enables AI systems to understand and navigate the organization’s knowledge.
“The ontology is at the heart of the information design of the AI powered enterprise. Without it, AI develops in a piecemeal and fragmented way,” Seth explains.
He illustrates this through a case study with Applied Materials, where technicians struggled to find answers across 14 disconnected systems. By developing a unified ontology and applying it to their data, the company cut search time in half and saved tens of millions of dollars annually.
How Organizations Can Begin
Seth outlines a practical process for building an ontology that supports scalable AI:
-
Identify the information bottlenecks that impede operations
-
Define solutions based on the root causes
-
Understand the specific use cases and user mental models
-
Establish the organizing principles that structure data consistently across systems
Once the ontology is in place, it becomes an evolving asset that supports future AI applications and ensures that new data remains aligned with business meaning.
A Path to Sustainable AI
Seth’s perspective is clear. AI cannot fulfill its promise without an information foundation that reflects how the business operates. Organizations that invest in ontology, governance, and structured knowledge will be positioned to achieve lasting value from AI. Those that skip this step will continue to face failure, rising costs, and fragmented systems.
Read the full Harvard Business Review article